0
$\begingroup$

I've been playing around with EEG data from an online database.

If the power line interference (PLI) is at 50Hz and sampling frequency is 64Hz, then according to the Nyquist theorem, the 50Hz PLI should be aliased with 14Hz.

But according to the database header file, the signal from EEG channel is processed by some prefilter, which is a bandpass filter with passband 0.5Hz - 32Hz.

So in the end, the raw EEG data I get from the database, is analog signal sampled at 64Hz and bandpass filtered in 0.5Hz - 32Hz range.

So my question is, do I have to design a notch filter to remove the PLI at 50Hz if the signal has already been bandpass filtered (32 Hz above removed.)

$\endgroup$
1
$\begingroup$

That depends if the bandpass filter was applied before or after sampling. If it was before sampling (i.e. in analog domain) it worked as an anti-aliasing filter and you dont need to filter out the PL. If it was applied in digital domain, you'd need to suppress it, because then the aliasing would occur. Have a look at the FFT of your signal to see if there is a peak or not at 14Hz, which could come from the PL.

(I guess the filter was applied in analog domain, because it does not make much sense to do it after sampling, since the high cutoff is already the Nyquist frequency.)

$\endgroup$
  • $\begingroup$ Thanks. I just checked the fft plot, and it turned out that no peak at 14 Hz, but there is a very high peak at 0 Hz... I believe the high pass prefilter of 0.5Hz is not working properly... $\endgroup$ – Tang Yun Mar 14 '17 at 5:58
  • $\begingroup$ So, if you want you can apply a DC-blocking filter in digital domain (or you just remove the mean of the signal). Does that answer your question? $\endgroup$ – Maximilian Matthé Mar 14 '17 at 6:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.